Skip to main content
ABC News
Market Forces Appear to Apply to Hospitals, Too

We tend to assume the market for health care works differently from, say, the market for refrigerators. When people buy a refrigerator, they compare product features and prices and make their decision. But when people are choosing a health care provider, it’s harder to compare “product” features. How good is one hospital compared to another when it comes to care? What are you really buying? And how can you find these things out?

As for prices, good luck finding any information on that. Besides, much of the time patients aren’t paying the full sticker price for a health care product or service anyway — an insurance company or the government is paying a big chunk of the final tab — so patients have less incentive to make cost-effective choices than a consumer who’s comparison-shopping for refrigerators might. And if patients can’t tell or don’t care whether they’re receiving high-quality, cost-effective care, then health care providers have fewer incentives than refrigerator manufactures to be more efficient.

This concept of “health care exceptionalism” — that the health care sector doesn’t operate according to standard market forces — has shaped much of the debate among policymakers about how to reform it. But how exceptional is it, really? Amitabh Chandra of Harvard University, Amy Finkelstein and Adam Sacarny of MIT and I set out to answer this question in a recent study. When we looked at the data, we discovered that the health care market might not be as exceptional as we think.

Using data on more than 3.5 million heart attack patients insured through Medicare spanning 1993 to 2007, we first measured how “efficient” individual hospitals in the U.S. were at treating these attacks. We decided to focus on heart attacks because cardiovascular disease is the country’s leading cause of death, so it’s an important area to study, and because the desired outcome, survival, is clear and easy to measure. Two-thirds of the patients in our sample were alive a year after their attacks. In addition, we wanted to study cases in which people didn’t have a lot of time to research their health care options. When they have more time — for instance, if they have a certain type of cancer — they’re more likely to choose facilities that are good at treating their particular illness, and a “selection” problem arises. Patients with the most challenging cases of an illness are more likely to seek or be sent for treatment at the best providers. But if a hospital’s performance is measured by its patient outcomes, top providers might not look as good as they really are because they’re likely treating some of the hardest cases. For heart attacks, this selection — of harder cases to better providers — is limited because of their acute nature.

We measured a hospital’s productivity by computing how long its heart attack patients lived after their episodes, relative to what one might expect given their demographics, health histories and treatment (for example, whether they had an angioplasty, received a stent or had bypass surgery). Using a regression model, we calculated how average survival across all patients in our sample was affected by age, race, gender, current and prior medical conditions, and the multitude of medical treatments the patients received. A hospital’s productivity was the average relative survival of its patients beyond, or short of, the level implied by the regression model.

We found some pretty big differences across hospitals in how efficiently they delivered services. Suppose a particular heart attack patient could receive the same medical procedures at different hospitals. Our measurements indicate that the patient would live on average more than 50 percent longer if he or she were treated at one of the more efficient hospitals than at one of the less efficient providers. For example, an older male patient suffering from other health issues, like diabetes, might survive eight months after receiving a bypass operation at one of the less efficient hospitals. If he received the same treatment at a better hospital, he might live an additional four months longer. We also found that our measure of hospital productivity correlated with quality metrics that Medicare calculated for hospitals’ heart attack treatment practices.

Additionally, we found that urban hospitals were 7 percent more efficient than rural hospitals — that is, their patients lived on average 7 percent longer, even after adjusting for the types of patients they treated and for the set of procedures the patients underwent. Teaching hospitals were 8 percent more efficient than non-teaching hospitals. And non-profit hospitals were 8 percent more efficient than for-profit hospitals, which were themselves 2 percent more efficient than government-run hospitals.

We then checked to see if the more productive hospitals — those delivering longer patient survival from a given set of medical procedures — received more patient traffic. They did. More efficient hospitals systematically treated more heart attack patients than did less efficient hospitals that were located in the same metro area. A hospital that was 10 percent more efficient than another hospital could expect to see 25 percent more patients. Furthermore, these better hospitals also grew faster and were less likely to close. That same 10 percent productivity advantage corresponded to a 1 percent faster growth rate in patient numbers over the following year, 4 percent faster growth over five years, and 6 percent faster growth over 10 years.

More than half of the heart attack patients in our study did not travel to the hospital nearest to where they lived. Instead, they went or were taken to more distant (and on average more efficient) hospitals for treatment. Mind you, this was while they were having a heart attack — not something people typically feel coming on and then sit down at the computer to do some research on their options. (Just imagine how they might respond if they had more time. The evidence here suggests that for other types of illnesses, like cancer, the rewards for top providers could be even greater.)

What explains the fact that patients were making their way to more efficient hospitals? It’s hard for us to say, unfortunately, since we don’t have data on what patients know and when they know it. They might learn from the experiences of family and friends, or perhaps doctors advise them during checkups. There’s even some indication that ambulance drivers, facing roughly the same drive time to two different hospitals, tend to choose the provider they know to be better.

We do know that patient traffic responds positively to a hospital’s patient survival performance and negatively to its overuse of inputs (like performing too many bypass surgeries in situations where stents would do just as well). Between these two factors, patient traffic is considerably more responsive to survival differences.

All this suggests that the health care market may not be as different as economists are prone to think. Even in settings where patients have little information about providers’ efficiencies, and insurance and time pressures reduce any incentive to find out, we see that patients still find their way to better hospitals. Something in the market is able to collect and transmit that performance information to patients or to someone who knows them.

That’s not saying the health care market is perfect. The next step is figuring out how patients learn about better providers and make decisions about their care. Anything we can do to make the health care sector even more responsive to efficiency differences across providers would be a good thing, especially with one-sixth of the country’s GDP, and our lives, on the line.

Chad Syverson is a professor of economics at the University of Chicago Booth School of Business.